From 4f3ccf9f96895888ed41b4d47faff0c059489a6c Mon Sep 17 00:00:00 2001
From: Isin Demirsahin <isin@google.com>
Date: Wed, 18 Dec 2024 15:31:17 -0800
Subject: [PATCH] Tests for Marathi diphthong grapheme -> long vowel phoneme
 rules.

PiperOrigin-RevId: 707687686
---
 nisaba/scripts/natural_translit/brahmic/g2p.py         |  4 ++++
 .../natural_translit/g2p/testdata/mr_iso_ipa.textproto | 10 ++++++++++
 2 files changed, 14 insertions(+)

diff --git a/nisaba/scripts/natural_translit/brahmic/g2p.py b/nisaba/scripts/natural_translit/brahmic/g2p.py
index 50be557a..9d10924a 100644
--- a/nisaba/scripts/natural_translit/brahmic/g2p.py
+++ b/nisaba/scripts/natural_translit/brahmic/g2p.py
@@ -39,6 +39,10 @@ def iso_to_txn() -> pyn.Fst:
 
 # Vowels
 
+# TODO: Convert this constant to a function where duration and diphthong
+# context are passed as arguments, and remove the recovery rule.
+# The current rule rewrites all /a/ to /ə/ including the diphthongs like /ai/,
+# and recovers long a with /ə:/ -> /a:/ but not the diphthongs.
 A_TO_EC = (
     rw.rewrite(ph.A, ph.EC) @
     rw.rewrite(ph.EC + ph.DURH, ph.A + ph.DURH)
diff --git a/nisaba/scripts/natural_translit/g2p/testdata/mr_iso_ipa.textproto b/nisaba/scripts/natural_translit/g2p/testdata/mr_iso_ipa.textproto
index 988e540b..962a2108 100644
--- a/nisaba/scripts/natural_translit/g2p/testdata/mr_iso_ipa.textproto
+++ b/nisaba/scripts/natural_translit/g2p/testdata/mr_iso_ipa.textproto
@@ -39,3 +39,13 @@ rewrite {
   input: "siddʰēgavhāṇa"
   output: "sid̪d̪ʰeːɡəʋʰaːɳ"
 }
+rewrite {
+  rule: "ISO_TO_IPA"
+  input: "aisā"
+  output: "ɛːsaː"
+}
+rewrite {
+  rule: "ISO_TO_IPA"
+  input: "kannauja"
+  output: "kən̪n̪ɔːd͡ʒ"
+}
\ No newline at end of file